M\'elange: Cost Efficient Large Language Model Serving by Exploiting GPU Heterogeneity.
Tyler GriggsXiaoxuan LiuJiaxiang YuDoyoung KimWei-Lin ChiangAlvin CheungIon StoicaPublished in: CoRR (2024)
Keyphrases
- language model
- cost efficient
- language modeling
- document retrieval
- n gram
- probabilistic model
- retrieval model
- speech recognition
- information retrieval
- smoothing methods
- test collection
- query expansion
- ad hoc information retrieval
- context sensitive
- statistical language models
- language modelling
- document ranking
- query terms
- relevance model
- word error rate
- query specific
- pseudo relevance feedback
- vector space model
- language model for information retrieval
- translation model
- document collections
- information extraction
- relevant documents